Add committor analysis about plumed2 HOT 5 CLOSED

I was thinking about implementing it. Can we consider it a form of "analysis" and implement it there?

from plumed2.

It should be simple to do it. Gareth added some time ago a way to stop the driver upon some condition. We have to modify the gromacs patch so as to do a clean stop when that happens and then add an Action to trigger it. I think it could go in analysis. Probably also PRINT and DUMPATOMS should go there, right?

from plumed2.

@carlocamilloni another thing: since for this you will have to change the gromacs patch, please do it directly on master branch. I changed on master branch the way patch are stored and modifying them should be easier - it will also allow merging with git several independent modifications to the patch.

from plumed2.

Hi Carlo,

Last week I was in Stuttgart and the guys from Espressomd suggested that
would be great to
have a general mechanism to let plumed pass back a vector that could be,
for example, an exit signal
that then can be recognized or used for some purposes (in gromacs for
dumping the checkpoint).
Then probably what is needed is a "condition" mechanism that handle this
condition. I just wanted to share this since I think it is appropriate
mechanism for the committor.
Ciao
D

Davide Branduardi
[email protected]
Web:
https://sites.google.com/site/davidebranduardi

On Wed, Oct 9, 2013 at 7:20 PM, Carlo Camilloni [email protected]:

I was thinking about implementing it. Can we consider it a form of
"analysis" and implement it there?

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/10#issuecomment-25990212
.

from plumed2.

Hi,

ok I have tested the possibilty of halting gromacs in a clean way. The code stops if the plumed stop flag becomes true, it writes the cpt files and close everything correctly (dumping performance statistics and so on). I will upload the new patch shortly and then I will start to work on the committor analysis.

Carlo

from plumed2.